The second edition of CODE_n’s new.New Festvial will be hosted in Stuttgart, also home of GFT’s Corporate Center. Naturally, we will once again take part, when startups and local companies come together to make innovation tangible. According to the motto “Intelligence X.0”, our team will focus on blockchain, cloud, data analytics and artificial intelligence – topics also seized by this years’ finalists. Like always, we have selected and spoken to the most interesting ones to present them on this blog, starting off with Cindicator from New York. If you are interested in artificial intelligence then read on! Mike Brusov, Co-founder and CEO of the startup, is taking it one step further – with hybrid intelligence.
On your website, you talk about hybrid intelligence for effective asset management. Could you explain your solution to us in more detail?
Mike: Yes, at Cindicator we’re working on Hybrid Intelligence for generating valuable predictive analytics for traders and investors. Hybrid Intelligence combines insights from 110,000+ analysts registered on our platform with several layers of artificial intelligence. Humans and machines are good at different things. As of today, only people can think creatively, but emotions and biases cloud our thinking. Artificial Intelligence (AI), or more precisely the set of different machine learning (ML) algorithms, can efficiently process vast volumes of data. We use this to remove biases from answers generated by the collective mind of thousands of analysts and utilize the wisdom of the crowds in a more efficient way.
We use Hybrid Intelligence to create predictive indicators for crypto and traditional assets. These indicators are available exclusively to holders of CND, our ERC-20 utility token for the internal digital economy. Our token holders use Hybrid Intelligence indicators in their trading and investment strategies.
What technology is your solution based on and how does it work?
Mike: The most important part of the Hybrid Intelligence ecosystem is our incentivised analyst community. Every month, analysts use our mobile and web apps to generate over 400,000 data points by forecasting different market events. They get points for correct answers and lose some for wrong answers. At the end of each month, they receive rewards as long as they’ve ended up with at least one point.
We collect this data from people and then apply over 50 different machine learning models to their answers. These models take into account many factors, including each analyst’s track record. We’re constantly experimenting with different Bayesian, Markov, weighting and linear models, as well as many others.
As a final layer, we use a neural network to identify complex, non-linear relationships between different models. Essentially, the neural network helps to pick the most effective ML model. We only launched the neural network in March and it has already increased its accuracy to 69%. By its very nature, neural networks “learn” as the volume of data increases, so we expect the accuracy will continue to increase. Neural networks are an active area of research for our data science team.
Whom is your solution for? What problems are you aiming to solve?
Mike: Our analytical products for both crypto and traditional markets are available exclusively to our token holders. About 5,000 of them regularly use our products. Some of them are individual traders who trade crypto assets part time. Some of them made it their job. Our products are also used by professionals and funds.
Regardless of who they are, all market participants face the same problem. Financial markets, especially crypto assets, are irrational, making it hard for investors and traders to make right decisions. Emotions, irrationality and a lack of data come into play. This uncertainty usually leads to substantial losses.
Do you think Artificial Intelligence will revolutionize the sector you work in? Maybe even our society?
Mike: Artificial Intelligence is already revolutionizing asset management in a lot of ways. Yet any AI application is impossible without high-quality data. To generate alpha, or an extra return that doesn’t rely on the overall market movement, this data should also be unique. If everybody has access to the same data, other market participants will quickly copy successful strategies. So, with Hybrid Intelligence we are creating both a unique data set (only for the participants of our digital ecosystem) and a cutting-edge machine learning stack to process it.
Yet as a technology Hybrid Intelligence is not limited to just asset management. I believe it would eventually be a valuable tool in any situation when you need to make a decision but don’t know what will happen in the future. Corporate R&D, scientific research, politics, and even sports are potential use cases. Effective decision making in uncertain times – is what we are aiming at.
Thank you for the interview, Mike!
If you want to find out more about GFT’s involvement at this year’s new.New Festival, have a look here.